Abstract There have been recent significant advances in understanding the role that immune-checkpoints play in down regulating the immune response to cancer. These discoveries have in turn led to the development of immune- checkpoint inhibitors that activate cytotoxic T-cells, and have demonstrated strikingly positive clinical outcomes across multiple tumor types. However, despite durable remissions in many patients, the overall response rate remains low. Immune checkpoint inhibitors are also associated with a high percentage of potentially lethal immune-related adverse events. Further, assessing therapeutic response is challenging, as tumors that may ultimately respond can appear to increase in size on anatomic imaging due to an influx of immune cells. This same immune infiltrate obscures FDG-PET analysis, as the immune cells are highly FDG avid. The lack of a useful response assessment has significantly complicated patient care and clinical development. Patients are frequently kept on therapies longer than necessary, as it cannot be ascertained whether they are responding. In order to address the difficulty with response assessment, there has been significant effort investigating predictive biomarkers, including novel imaging methods. The imaging biomarkers analyzed thus far have focused on identifying the presence of tumoral immune infiltrate and have not proven strongly predictive of response. Their lack of utility is likely because they cannot distinguish between active and inactive immune infiltrate, the latter of which is hypothesized to be a common cause of immunotherapy failure. To monitor cytotoxic T lymphocyte (CTL) activity, we have developed a first-in-class peptide-based PET imaging agent that binds to granzyme B, a serine protease released by CTLs when they are actively attacking tumor cells. We have demonstrated our imaging agent in two different immunotherapy models and shown that it is able to predict response to checkpoint inhibitors. We have also interrogated checkpoint-inhibitor treated human melanoma samples for granzyme B expression. These results corroborate our pre-clinical findings of high granzyme B expression correlating with response to immunotherapy. Finally, we designed a human analogue of our peptide, which specifically bound to granzyme B in human samples. This proposal aims to finalize an optimized human probe and inform the patient population and timing for near-term clinical evaluation. To achieve this goal, we will first develop second-generation peptides that may provide enhanced affinity or improved pharmacokinetics for granzyme B measurement, and assess them in humanized mouse immunotherapy models. In order to better structure clinical trial imaging time-points, we will continue our assessment of granzyme B expression in human checkpoint inhibitor treated melanoma biopsy specimens. Quantification of target expression focused on dosing intervals will help to maximize clinical impact by identifying response prior to administration of subsequent therapy. Together, we hope these aims can rapidly advance granzyme B imaging into the clinic to provide the response biomarker that is so desperately needed.